President Trump has signed an executive order establishing a voluntary framework for reviewing advanced artificial intelligence models before public release. Titled "Promoting Advanced Artificial Intelligence Innovation and Security," the directive instructs federal agencies to create a classified benchmarking process for systems with significant cybersecurity capabilities.
The National Security Agency will lead this effort alongside the Treasury Department, Department of Defense, and Department of Homeland Security. Developers are encouraged to provide access to covered frontier models thirty days prior to launch for security evaluations. This window was reduced from an initial ninety-day proposal to maintain U.S. competitiveness against China.
Crucially, this framework replaces previous administration compliance requirements with a voluntary system. There are no mandatory licensing regimes or new bureaucratic hurdles imposed on developers. The policy aims to establish uniform federal standards while avoiding stifling innovation in a sector where speed determines market valuation.
Market participants should note potential minor delays in frontier model releases as labs navigate this new security partnership. Decentralized AI projects currently remain in a regulatory gray area due to their lack of centralized corporate entities. The administration retains flexibility to adjust timelines should geopolitical competitive pressures intensify.